Mutagenesis of the N- and C-terminal cysteine pairs of Tn501 mercuric

X-ray Structure of a Hg Complex of Mercuric Reductase (MerA) and Quantum ... Lyase MerB: Catalysis, Product Release, and Direct Transfer to MerA. Greg...
0 downloads 0 Views 3MB Size
Download PDF
Biochemistry 1989, 28, 1183-1 194

Walsh, C., Fisher, J., Spencer, R., Graham, D. W., Ashton, W. T., Brown, J. E., Brown, R. D., & Rogers, E. F. (1978) Biochemistry 17, 1942. Whitesides, G., & San Filippo, J. (1970) J . Am. Chem. SOC. 92, 661 1-24. Wilkinson, K., & Williams, C., Jr. (1979a) J. Biol. Chem. 254, 852-62. Wilkinson, K., & Williams, C., Jr. (1979b) J . Biol. Chem. 254, 862-7 1.

1183

Williams, C. H . (1976) Enzymes (3rd Ed.) 13, 89. Williams, C. H., Arscott, D. L., Matthews, R. G., Thorpe, C., & Wilkinson, K. D. (1979) Methods Enzymol. 6 2 0 , 185. Williams, C. H., Arscott, D. L., & Schulz, G. (1982) Proc. Natl. Acad. Sci. U.S.A. 79, 2199. Wilson, G. S . (1978) Methods Enzymol. 54, 396. Zoller, M. J., & Smith, M. (1986) Methods Enzymol. 100, 468. Zoller, M. J., & Smith, M. (1984) DNA 3, 479.

Mutagenesis of the N- and C-Terminal Cysteine Pairs of Tn501 Mercuric Ion Reductase: Consequences for Bacterial Detoxification of Mercurialst Melissa J. Moore and Christopher T. Walsh* Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, Massachusetts 021 15 Received June 1, 1988; Revised Manuscript Received September 20, 1988

ABSTRACT: Mercuric ion reductase (the merA gene product) is a unique member of the class of FAD and redox-active disulfide-containing oxidoreductases by virtue of its ability to reduce Hg(I1) to Hg(0) as the last step in bacterial detoxification of mercurials. In addition to the active site redox-active disulfide, formed between Cys135and Cys14 in Tn501 MerA, the protein products of the three merA gene sequences published to date have two additional conserved pairs of cysteines, one near the “terminus (CysloCys13in Tn501 MerA) and another near the C-terminus (cys558cys559 in Tn501 MerA). Neither of these pairs is found in other members of this enzyme family. To assess the possible roles of these peripheral cysteines in the Hg(I1) detoxification pathway, we have constructed and characterized one single mutant, Cysl&la13, and two double mutants, AlaloAla13and Ala558Ala559.The N-terminal mutants are fully functional in vivo as determined by HgC12 resistance studies, showing the N-terminal cysteine pair to be dispensable. In contrast, the Ala558Ala559 mutant is defective for HgC12 resistance in vivo and Hg(SR)2 reduction in vitro, thereby implicating CyssSs and/or CysSs9in Hg(I1) reduction by the wild-type enzyme. Other activities, such as NADPH/thio-NADP’ transhydrogenation, N A D P H oxidation, and DTNB reduction, are unimpaired in this mutant.

M e r c u r i c reductase is the flavoenzyme that catalyzes the last and crucial step in Hg(I1) detoxification by bacteria [for review see Foster (1983) and Summers (1986)l:’ Hg(SR),

+ NADPH + H+

-

Hg(0)

+ NADP’ + 2HSR

It shows strong active site similarity to the disulfide oxidoreductases, pig heart lipoamide dehydrogenase, human glutathione reductase, and trypanothione reductase (Fox & Walsh, 1983; Shames et al., 1986), which catalyze the flow of electrons from NAD(P)H to disulfide substrates via a flavin coenzyme and an active site redox-active disulfide. However, while the three former proteins are all capable of binding mercurials in their two-electron-reduced states (Massey & Williams, 1965; Casola & Massey, 1966; Miller et al., 1986), only mercuric reductase is able to catalyze reduction of Hg(I1) at a useful rate. In contrast, the physiological activities of both glutathione reductase and lipoamide dehydrogenase are potently inhibited by mercurials. TnJOl mercuric reductase is encoded by the merA gene of the mer operon in transposon Tn501, originally isolated on a plasmid from Pseudomonas aeruginosa (Stanisich et al., 1977). This operon is composed of a regulatory gene, merR, ‘This work was supported in part by NIH Grant GM 216439. M.J.M. was the recipient of a National Science Foundation Predoctoral Fellowship. * To whom correspondence should be addressed.

0006-2960/89/0428-1183%01.50/0

Table I: Primary-Sequence Cysteine Pairs in the TnSOl mer Operon MerR“ 113-LeuVal@Ala@HisAla-119 MerTb 21-AlaSerAlaCysCysLeuGly-27

76-@LysProGlyGluValC~g-82 MerPb MerA‘

3 1 -MetThr&sAlaAla&sPro-37 8-MetThr@AspSer@Ala- 14 1 3 4 - T h r ~ V a l A s n V a l G l y140 ~555-GlnLeuSerCysCysAlaGly-56 1

“OHalloran & Walsh, 1987. bMisra et al., 1984. CBrownet al., 1983.

an operator/promoter region, and four structural genes: mer T, -P, -A, and -D. merT and merP encode Hg(I1)-transport and periplasmic Hg(I1)-binding proteins, respectively, while no definitive function has yet been assigned to merD. A recurring structural motif throughout these mer gene products is the Abbreviations: NADPH, reduced nicotinamide adenine dinucleotide phosphate; thio-NADP*, thionicotinamide adenine dinucleotide phosphate; FAD, flavin adenine dinucleotide; DTNB, 5,5’-dithiobis(2-nitrobenzoate); TNB, 5-thio-2-nitrobenzoate dianion; EDTA, ethylenediaminetetraacetic acid; IPTG, isopropyl /3-o-thiogalactopyranoside;bp, DNA base pair(s); SDS, sodium dcdecyl sulfate; PAGE, polyacrylamide gel electrophoresis;HPLC, high-performanceliquid chromatography;N or C terminal, proximal to the amino or carboxyl terminus of a protein; E,,, EH,, and EH4, oxidized, two-electron-reduced, and four-electronreduced forms, respectively, of mercuric reductase or of the disulfide oxidoreductases; KSi,inhibition constant for binding of a second substrate molecule to the enzyme-substrate complex.

0 1989 American Chemical Society

1184 Biochemistry, Vol. 28, No. 3, 1989 primary sequence grouping of paired cysteines (Table I), a pair being defined as two cysteines separated by no more than five other amino acid residues. Because of the affinity of Hg(I1) for bis-thiol ligation, it has been suggested that these cysteine pairs may play key roles in the detoxification pathway. The roles proposed include initial binding of Hg(I1) in the periplasm, its transport through the cell membrane via specific transfer from one cysteine pair to the next, and its correct positioning in the active site of mercuric reductase for catalytic reduction (Brown, 1985). Mercuric reductase itself has eight cysteines, six of which are paired. It is these three cysteine pairs, which are all conserved in the three mercuric reductases whose sequences have been published (Brown et al., 1983; Misra et al., 1985; Laddaga et al., 1987), that we have chosen as targets for site-directed mutagenesis. Using this method, we have previously shown that the Cys135Cys140 pair, which forms the active site redox-active disulfide of mercuric reductase, is essential for normal catalytic Hg(I1) reduction (Schultz et al., 1985). The preceding paper (Distefano et al., 1989) describes further mutagenesis of the redox-active pair and the mechanistic implications of said mutants’ behavior. In this paper we report the directed mutagenesis of the Nand C-terminal cysteine pairs at positions 10, 13 and 558, 559, respectively. The most significant result of this work is the identification of the first structural element unique to mercuric reductase, Le., unique among the disulfide oxidoreductases, that is essential for high levels of Hg(I1) reductase activity. We report herein the in vivo and in vitro characterization of a double mutant of the C-terminal cysteine pair, Ala558Ala559. This mutant is catalytically impaired in Hg(SR)2 reduction, yet retains wild-type levels of all other activities assayed, including NADPH/thio-NADP’ transhydrogenation and aryl disulfide (DTNB) reduction. In the following paper (Miller et al., 1989) we build on these results to show that wild-type mercuric reductase actually has four cysteinyl thiols in its active site: cys135,cys140,cys558, and cysSs9. We also demonstrate that Cys558 and Cys559are capable of forming a second, auxiliary disulfide whose reduction is a necessary priming step for optimal enzyme activity. While the C-terminal cysteines are essential for Hg(I1) reduction, it has previously been shown by in vitro proteolysis that the 85 amino acid, N-terminal domain of mercuric reductase is not involved in catalysis (Fox & Walsh, 1983). This, coupled with its homology to the merP gene product, led Brown (1 985) to suggest that the N-terminal domain plays a role in intracellular Hg(I1) transport via the CysloCys13 thiols. However, by in vivo characterization of both a single mutant, CysloAla13,and a double mutant, AlaloAla13,of this pair we show here that these cysteines are not essential for Hg(I1) detoxification. We also present the construction of a shortened form of the merA gene, merAA85, that should encode an enzyme that is the wild-type mercuric reductase minus its 85-residue N-terminal domain. Unfortunately, merAA.85 fails to produce significant levels of protein, and thus the function of the N-terminal domain, if any, remains a mystery. EXPERIMENTAL PROCEDURES Materials Escherichia coli strain CSH26 recA containing plasmid pUB345 1 (a p J O E l l 4 derivative with BamHI linkers in the 581 1 bp Not1 site) was the generous gift of Dr. Nigel Brown of the Department of Biochemistry, University of Bristol, Bristol BS8 lTD, UK. E . coli strain JM101, which was used for mutagenesis and routine M 13 manipulations, and M 13 R F

Moore and Walsh DNAs were obtained from New England Biolabs. E . coli strain RZ1032 (Kunkel, 1988), which was used as the dutung- strain in later M13 mutageneses, was the kind gift of Dr. Thomas A. Kunkel. E . coli strain HB101*, a derivative of HB101 giving higher transformation frequencies, was used for all plasmid manipulations and was kindly provided by Dr. David Botstein, Department of Biology, MIT, Cambridge, MA. Plasmids pJOE114, pSE181, and pPSO1, M13 derivative M13ps1, and E . coli strain W3110 lacIq were described previously (Schultz et al., 1985). Yeast glutathione reductase, Pseudomonas cepacia protocatechuate reductase, and milk xanthine oxidase were gifts from Drs. Charles Williams, Jr., David Ballou, and Vincent Massey, respectively. All restriction enzymes, T4 polynucleotide kinase, T4 DNA ligase, and DNA polymerase I large fragment (Klenow) for mutagenesis and plasmid constructions were obtained from New England Biolabs, calf intestinal phosphatase was from Boehringer-Mannheim, and avian reverse transcriptase was from Pharmacia. Unlabeled reagents and Klenow fragment for Sanger sequencing were from Amersham or BoehringerMannheim, while 32P-and 35S-labeled nucleotides and 35Slabeled amino acids were purchased from Amersham or New England Nuclear. Protected mononucleotides for phosphotriester oligonucleotide synthesis were obtained from Biosearch. Orange A Matrex gel was purchased from Amicon Corp. and P6-DG desalting gel from Bio-Rad. All other reagents and chemicals were of the highest grade commercially available. Methods DNA Manipulations. All techniques used in M13 template construction, preparation, and Sanger sequencing are described in the M I 3 Cloning and Sequencing Handbook (Amersham, 1983) with the exception that uracil-containing M13 templates were prepared as described by Kunkel (1985). Transformations were performed according to the simple transformation protocol of Hanahan (1985). Reverse transcriptase sequencing of double-stranded plasmid DNA was as described by Seidmann (1986). All other cloning techniques and the MaxamGilbert sequencing protocol can be found in Molecular Cloning: A Laboratory Manual (Maniatis et al., 1982). M I 3 Templates. The M13 template used for directed mutagenesis in the N-terminal portion (Le., residues 1-186) of merA, M13ps1, has been described elsewhere (Schultz et al., 1985). A template containing the C-terminal region (Le., residues 392-561) of merA, M13mm1, was constructed by inserting the 775 bp SphIIBamHI fragment of pUB345 1 into SphIIBamHI-digested M13mp19 (see Figure 2). This insertion was confirmed by Sanger sequencing with universal primer. Oligonucleotides. Oligonucleotides for both mutagenesis and sequencing were synthesized according to a modified phosphotriester procedure on a Biosearch Sam1 automated DNA synthesizer and were purified by polyacrylamide gel electrophoresis (Biosearch, 1984). Mutagenesis. ( A ) CysloAla13 and AlaloAla13. The CysloAla13mutation was generated in the M13 vector, M13ps1, with a 17-mer oligonucleotide, 5’CGCCGCGGCCGAGTCGC-3’. M13psl containing this mutation, M13mmal3, was subsequently used as the template for Cysl&la13 to Alal&la13 mutagenesis via a second 17-mer, generating 5 ’-C G A G T C G G C A G T C A T G C - 3 ’, M13mmalOal3. Both primers contained two mismatched base pairs (italicized), and a primer:template ratio of 30:l was used in both mutagenesis mixtures. Mutagenesis reactions were carried out by using a two-primer modification of the standard Zoller and Smith (1983) procedure described by

N- and C-Terminal Cys Pairs of Mercuric Reductase

Biochemistry, Vol. 28. No, 3, 1989 1185

Hind

[

Sal1I M13mma13

HindllllSall __c

I SacllMindlll

Narl

I

Hindlll

HindllllSacll

FIGURE 1:

Construction of pJOEI 14 varients containing the Ala,, (pMMal3) and Ala,&Ia,, (pMMalOal3) mutations in “ A .

Messing (l983), except that the alkaline sucrose gradient centrifugation was omitted. Mutations were initially detected via hybridization screening of first-round phage isolates, followed by Sanger sequencing after a second round of phage growth. By use of the above techniques the Cysl&lal, and AlaloAlal, mutations were obtained with 0.8% and 1.7% efficiencies, respectively. Unfortunately, in addition to the generation of the desired mutations in low yields, these mutageneses were also plagued with both spurious deletions of insert DNA and the production of undesired point mutations. The CysloCysl, to Cysl&lal, mutagenesis reaction produced an additional point mutation in the merT gene, while the Cysl&lal, to Alal&lal, mutagenesis yielded a 350 bp deletion in merA immediately 3‘ to the Alal, codon. The formation of these secondary mutations was probably exacerbated by both the large size of this M13 insert (1748 hp) and the high GC content [approximately 65% (Brown et al., 1983)] of the TnSOI mer operon. In order to remove these secondary mutations, both the Cysl&lal, and AlaloAlal, mutations were reconstructed into the mer operon of p J O E l l 4 via the series of restriction fragment replacements described below (Figure 1). (B) M13mmA85 and M13mma558a559. To increase mutagenesis efficiencies, both subsequent mutageneses (M13psl to M13mmA85 and M13mml to M13mma558a559) were carried out on templates that had been grown for two generations in the dut- ung-strain RZ1032 (Kunkel, 1985, 1988). M13mmA85, in which merA residues 2-86 have been deleted, was generated from M13psl by using a 2l-mer. 5’GGCGGCGGCCATGCAACAGAT-3’,that primed at both ends of the desired deletion, causing the intervening DNA sequence to be looped out and lost. M13mma558a559 contains the Alass,Alass, double mutation and was generated in one step from M 1 3 m m l with a 20-mer, 5’CCCGGCGGCGGCGGAAAGCT-3‘. possessing four mis-

matched base pairs (italicized). Optimal primer:template ratios were determined to he 5:l for the former primer with M13psl and 1 0 1 for the latter with M13mml by using the specific priming test via Sanger sequencing described by Zoller and Smith (1983). Mutagenesis reactions were performed as described by Kunkel (1985) except that uracil glycosylase treatment was omitted. These mutagenesis mixtures were used to transform JM101, and phage prepared from resulting plaques were screened for mutations directly by Sanger sequencing. According to this method the mutant M13 derivatives, M13mmA85 and M13mma558a559, were obtained in 12.5% and 16.7% yields, respectively. N o secondary mutations were found to result from either of these mutageneses. Plasmid Constructions. (A) pMMal3. The intact Cys,oAla,, merA gene was reconstructed in the mer operon (Figure 1) by first replacing the 1748 bp HindIII/SalI fragment of pJOE114 with that of M13mma13. However, this plasmid, pMM02, contains an additional undesired mutation in merT (see above). Therefore, a pJOEl14 derivative containing only the CysloAlal, merA mutation, pMMal3, was constructed by excising the merT-containing NarI fragment of pMMO2 and replacing it with the same fragment from wild-type pJOEl14. (B) pMMaIOal3. The Ala,&laI3 merA gene was reconstructed by taking advantage of a Sac11 site created by the Ala,, mutation. The 1236 bp HindIIIISacII fragment of M13mmalOa13 was cloned into HindIII/SacII-digested pMM02. This was followed by replacement of the 3280 bp NarI fragment with that of p J O E l l 4 as described above to create pMMalOal3 (Figure I). These manipulations ensured that only the 584 bp NarIISalI fragments of p M M a l 3 and pMMalOal3 were ever present as singlestranded DNA that had undergone primer-mediated mutagenesis. To eliminate the possibility of any remaining secondary mutations, the sequences 50 bp to either side of the desired mutations were confirmed by double-stranded MaxamCilbert sequencing

1186 Biochemistry, Vol. 28, No. 3, 1989

Moore and Walsh

I

SphlIBsmHl

Construction of MI 3mml (C-terminal merA mutagenesis template), pMMa558a559 (pUB345 1 containing the Ala5,8Ala55y mutation in merA), and the mutant MerA overproducing plasmids, pMMOa558a559 and pMMOalOal3. pMMOalOal3 construction and its insert into pSE181 are shown in parentheses. FIGURE 2:

from the NarI site a t bp position 3839 of both plasmids. Additionally, the 566 bp XhoII/SalI fragment of p M M a l 3 was subcloned into BamHIISalI-digested M13mp18 (not pictured) and Sanger sequenced in its entirety. (The same fragment of pMMalOal3 was derived from pMMO2, the precursor to pMMal3, during plasmid reconstruction as described above, eliminating the necessity to similarly resequence it.) (0pMMA85 and pMMa558a559. pMMA85 was constructed by inserting the 329 bp NarI/SalI fragment of M13mmA85 into the 11 490 bp SalIINarI fragment of pJOE114 generated by complete digestion with Sal1 and partial digestion with NarI (not pictured). pMMa558a559 was constructed by ligating the 775 bp SphIIBamHI fragment of M13mma558a559 with the 11 299 bp pUB3451 fragment generated by complete digestion with BamHI and partial digestion with SphI (Figure 2). Both pMMA85 and pMMa558a559 were confirmed by restriction mapping and MaxamGilbert or reverse transcriptase sequencing in the area of the desired mutation, and as above, the XhoIIjSalI pMMA85 fragment and the SphIIBamHI pMMa558a559 fragment were subcloned back into M I 3 and completely resequenced (not pictured). ( D ) pMMOa558a559 and pMMOalOal3. Tac-based overproducers of the Ala558Ala559and Ala,,Ala,, mutant mercuric reductases were constructed by cloning the 1861 bp Aha11 fragment of pMMa558a559 or pMMalOal3, respectively, into ClaI-digested pSE181 (Figure 2). These constructions, pMMOa558a559 and pMMOalOal3, were screened for insert directionality by restriction mapping and for their ability to produce bands of the correct molecular mass by SDS-PAGE after induction with IPTG in strain W31 IO laclq. pMMOa558a559 was used to overproduce AlaS,8AlaS5y mercuric reductase as described below, while pMMOalOal3

was used to generate Ala,,Ala,, enzyme as presented in the following paper (Miller et al., 1989). Growth Curves. Klett units of turbidity were measured on a Klett colorimeter with a No. 808 red filter. HBIOI* cells containing the appropriate plasmid were cultured overnight a t 37 OC in amp' 2 X Y T medium. These cultures were diluted 1:lOO into amp' 2 X Y T medium and grown to approximately 35 Klett units. The cultures were then divided, varying concentrations of HgCIz were added, and growth was monitored by taking Klett measurements every 30 min for a total of 7I/, h (Nakahara et al., 1979b). Pulse Labeling of HgCI,-Induced Proteins. HBIOI* cells transformed with either pMMA85 or pMMa558a559 were cultured in supplemented M9 minimal medium (20 pg/mL thiamine hydrochloride, 0.1% each of the L-amino acids, excluding cysteine and methionine). At A,,, = 0.43, the cultures were divided in half and 0.5 p M HgCIz was added to one half to induce mer operon translation. After a 15-min incubation period, [35S]-~-methionineand ['5S]-~-cysteinewere added to both induced and uninduced cultures to a final concentration of IO pCi/mL each. One-milliliter samples were removed at 2, 5, IO, and 20 min after the addition of the radioactivity. The samples were quickly centrifuged, and the cell pellets were flash-frozen in dry ice/ethanol and stored a t -70 "C. Total cellular protein was electrophoresed on a 5-20% polyacrylamide gradient SDS gel (Hames, 1981) and the dried gel exposed to film overnight. Autoradiographs were scanned with an LKB UltroScan XL laser densitometer. Protein Purification. Both the wild-type and Ala558Ala,5y enzymes were purified from W3110 lacIq/(pPSOl or pMMOa558a559) cells as previously described (Schultz et al., 1985). with the following modifications: (a) the cells were induced at = 0.7 by the addition of only 1 mM IPTG and were harvested 4'/,-6 h thereafter; (h) no protease inhibitors were added to the purification buffer; (c) after enzyme elution from the Orange A column with 200 p M NADPH, the p l e d fractions were incubated at 4 "C with excess FAD to ensure maximal loading of flavin. Excess FAD and NADPH were removed by passing samples through a Bio-Rad P6-DG desalting column. Enzyme samples were routinely stored a t 4 'C in 20 mM sodium phosphate (pH 7 . 9 , 0.5 m M EDTA, and 0.1% p-mercaptoethanol or at -70 O C after addition of glycerol to 15% (v/v). Enzymebound N A D P was removed via exhaustive dialysis of enzyme against 2 M KBr (Fox & Walsh, 1982) in 20 mM sodium phosphate (pH 7.5) and 0.5 mM EDTA, followed by incubation with excess FAD and desalting on a P-6DG column as above, using the same buffer minus KBr. 8-Mercaptoethanol was not added to KBr-dialyzed samples. Spectrophotometry. UV/visible spectra and single-wavelength measurements were recorded on a Perkin-Elmer 554, a Perkin-Elmer Lambda 5 double-beam, or a Hewlett-Packard 8452A diode array spectrophotometer. Fluorescence spectra were recorded at 2-4 OC by Dr. Vincent Massey, Department of Biological Chemistry, University of Michigan, Ann Arbor, MI. Other fluorescence measurements were made with a Perkin-Elmer LS-3 fluorometer at ambient temperature. Enzyme Concentrations. For activity measurements and thiol titrations, enzyme concentrations were based on flavin content, using an extinction coefficient of 11.3 mM-' cm-I at 456 nm for both the wild-type and Ala,,,Ala,,, enzymes. Thiol titrations were only performed on enzyme samples with an A272/AIS6 spectral ratio less than 6.5, ensuring that the flavin concentration truly reflected the concentration of protein monomer.

N- and C-Terminal Cys Pairs of Mercuric Reductase Physical Characterization. The extinction coefficient for flavin in the Ala5S8AlaS59 mutant was determined by liberating FAD from the enzyme with heat in 10 mM MgC12 as described by Schultz et al. (1985). Monomer protein molecular masses were established by SDS-PAGE, while native molecular masses were determined by gel filtration on a Waters HPLC equipped with a Bio-Rad 300 mm X 7.5 mm TSK 250 column and 75 mm X 7.5 mm TSK 250 guard column. Samples were run in 50 mM sodium sulfate and 20 mM sodium phosphate, pH 6.8, at 1.0 mL/min. Thiol Titrations. P-Mercaptoethanol was removed from enzyme samples to be used for thiol titrations via a Bio-Rad P6-DG desalting column. Thiols were titrated in aerobic solutions with DTNB in 4-5 M guanidine hydrochloride as described previously by Fox and Walsh (1982) or in 6 M guanidine hydrochloride by the method of Riddles et al. (1983). Reduced enzyme samples were generated by addition of 200 pM NADPH (25 "C) and were then denatured with guanidine hydrochloride within 1 min of this addition. An extinction coefficient of 13.7 mM-' cm-' at 412 nm was used for TNB in 6 M guanidine hydrochloride (Riddles et al., 1983). Anaerobic Redox Titrations. Anaerobic enzyme titrations with NADPH and dithionite were performed as described elsewhere (Miller et al., 1989; Fox & Walsh, 1982) in double-side-armed anaerobic cuvettes similar to those of Williams et al. (1979). Enzyme samples were made anaerobic by repeated cycles of evacuation and flushing with argon that had been passed through an oxygen-absorbing cartridge. Dithionite solutions were standardized anaerobically against riboflavin both before and after each set of enzyme titrations. Redox Potentials. Midpoint potentials for the E,,/EH2 and EH2/EH4 enzyme couples were determined by spectrophotometrically following the reduction of KBr-dialyzed enzyme, with a coupled xanthine/xanthine oxidase system as the reductant, in the presence of an appropriate indicator redox dye. (NADP+-free enzyme samples are used to avoid unnecessary complications from both the free and enzyme-bound NADP+/NADPH couples.) This method will be described in detail elsewhere [V. Massey et al., unpublished results; see also footnote 2, Distefano et al. (1989)l. The Ala5s8Ala5s9 E,,/EH2 potential was established in 100 mM sodium phosphate (pH 7.5) with the redox dye 1-deazariboflavin [ E O ' = -280 mV (Walsh et al., 1979)], while its EH2/EH4 potential was determined with benzylviologen [ E O ' = -358 mV (Wilson, 1978)] in 100 mM sodium phosphate (pH 7.0). Concentrations of oxidized and reduced species and the midpoint potentials were calculated in the same manner as that described by Fox and Walsh (1982). Enzyme Assays. ( A ) Aerobic. Hg(SR)2-independent and Hg(SR),-dependent NADPH oxidation assays were performed aerobically at 37 OC in 80 mM sodium phosphate (pH 7.4), 200 pM NADPH, and 1 mM P-mercaptoethanol. NADPH consumption was monitored at 340 nm (absorbance), and O2 consumption rates were measured with a Yellow Springs Instrument Co. biological oxygen monitor (Model 53). ThioNADP+-dependent NADPH oxidation was monitored at ambient temperature in 80 mM sodium phosphate (pH 7.4), 1 mM P-mercaptoethanol, and 100 pM thio-NADP' by either NADPH absorbance at 340 nm or fluorescence emission at 470 nm (excitation at 340 nm). Transhydrogenase rates in crude extracts were measured similarly with 200 pM NADPH at 25 OC. ( B ) Anaerobic. DTNB reduction rates were determined in anaerobic solutions that were prepared in a helium atmosphere glovebox and then sealed in stoppered test tubes. TNB pro-

Biochemistry, Vol. 28, No. 3, I989

1187

duction was monitored at 412 nm (e = 13.7 mM-' cm-') or at 440 nm ( E = 9.6 mM-' cm-') with assays carried out in 80 mM sodium phosphate (pH 7.4) and 200 pM NADPH (25 "C). 203Hg(0)volatilization assays were carried out in 10-mL Wheaton vials as described by Distefano et al. (1989) except that reaction mixtures contained 100 pM HgC12, 1 mM cysteine, and 400 pM NADPH in 100 mM sodium phosphate (pH 7.4). To ensure anaerobiosis, assay mixtures containing all components (except 203HgC12)were incubated for 1 h (37 "C) with protocatechuate and protocatechuate dioxygenase. Reactions were then initiated by adding anaerobic 203HgC12, and its volatilization was monitored as described. The specific activity of the yeast glutathione reductase sample used for ,03Hg(0) volatilizations was checked according to the conditions given by Massey and Williams (1965). RESULTS Growth Curves. The three mutations in the N-terminus of Tn501 mercuric reductase (CysloAla13,AlaloAla13, and merAA85) were created to test the hypothesis that the Nterminal domain is involved in Hg(I1) transport via the Cyslo and Cys13thiols. However, since none of these mutations was expected to affect the enzyme's catalytic properties (see Discussion), the consequences of each (as well as the Ala558Ala559 mutation) were first assessed by examining its in vivo effect on the HgC12 resistance phenotype. Resistance phenotypes were determined by challenging HB101* cells containing each mutant mer operon with various concentrationsof HgC12and following their subsequent growth in liquid medium (Nakahara et al., 1979b). As can be seen in Figure 3, this method allows the distinction of three main phenotypic classes: HgC12 sensitive (Figure 3A), HgC12 supersensitive (Figure 3B), and HgC12 resistant (Figure 3C). HgC12-sensitive cells are those containing an ampicillin resistance determinant but no mer genes. These cells are able to grow in up to 25 pM HgC12 but are killed by 50 pM HgC12. Resistant cells contain a functionally intact mer operon. They show some slowing of growth at all HgC12 concentrations tested, probably due to amplified mer protein production from this multicopy plasmid system, but grow well even at 250 pM HgC12. Supersensitive cells are those that retain the specific Hg(I1)-transport system but lack a functional mercuric reductase. These cells essentially commit suicide by concentrating Hg(I1) in their cytoplasms and characteristically exhibit growth inhibition in HgC12 concentrations as low as 2.5 pM. Conversely, cells that have a working mercuric reductase, but are impaired in Hg(I1) transport, have phenotypes very similar to Hg(I1)-sensitive cells (not shown). Under the conditions of this assay both pMMa13 (encoding Cysl&la13 mercuric reductase) and pMMalOal3 (Alal&la13 mercuric reductase) are indistinguishable from wild-type pJOE114 in their conferred resistance (Figure 3C). This result indicates that at least in this gene-amplified system (see Discussion) neither Cyslo nor Cys13is essential for Hg(I1) transport as proposed. In sharp contrast, however, cells carrying either pMMA85 (encoding N-terminus-deleted mercuric reductase) or pMMa558a559 (Ala558Alas59 mercuric reductase) exhibit a HgC12-supersensitivephenotype (Figure 3B). HgCl,-Inducible Gene Products of pMMA85 and pMMa558a559. The observation that both pMMA85 and pMMa558a559 confer supersensitivity to HgC12 suggests that neither mutant merA gene encodes a catalytically functional mercuric reductase. Correspondingly, no Hg(SR),-dependent consumption of NADPH was observed in HBlOl pMMA85 crude cell extracts after induction with subtoxic HgC12 con-

Biochemistry, Yo/. 28, No. 3, 1989

1188

Moore and Walsh

A. SENSITIVE

C. RESISTANT

B. SUPERSENSITIVE 0.0.d

200

ouu05sy)0ss9

--g .-. -

-r f

100

100

: 0

0

2

4

1

8

6

,

4

1

1

6

1

1

8

FIGURE 3 Growth curve assays for HgCl,-%nsitive (A), -supersensitive ( 6 ) or -resistant (C) phenotypes. HBIOI. cells containing the indicated plasmids were grown at 37 OC in 2 X YT liquid medium to approximately 35 Klett units of turbidity, when the cultures were divided and the indicated concentrations of HgCI, added (arrows). In (C) HgCI, was first added to 50 r M , and then after 1 h of induction the concentration was boosted to 250 r M total (see Discussion). Controls contained no HgCI2. pKK223 has no mer genes, while pJOEl14 contains the wild-type TnSO1 mer opron. pMMa558a559, pMMA85. pMMal3, and pMMalOal3 are pJOEl14 variants e n d i n g the mutant mercuric reductases described in the text.

HgCI,

minutes

'. 2

DMMA 85

DMMa558a559

. . + + + +

5 10 20

2

5 10 20

... . 2

5 10 20

+ + + *1 2

5 10 20

-native -clipped

MerA MerA

*I**...

E - - -

-

I

,nascent -processed

MerP Mer?

Pulse labeling of HgC12-inducedproteins e n d e d by pMMa558a559 and pMMA85. Cultures were grown (see Methods) to A,, r M HgCIz was added to half of each. After 15 min, ["Slcysteine and ['JS]methionine were each added to IO aCi/mL, and I-mL samples were removed at 2, 5, IO. and 20 min thereafter. Cells were pelleted and flash frozen to halt metabolism, and then total cell protein was electrophoresed on a denaturing 5-20% polyacrylamide gradient gel. Presence or absence of HgCI, induction is indicated by (+) and (-), while the bars indicate the molecular weights of expected protein products. FlGuRe 4

= 0.43, and 0.5

centrations (data not shown). For each mutant there are four possible scenarios that would result in the above observations: ( I ) the mutant merA gene is not effectively translated, (2) the polypeptide is synthesized but does not fold properly, (3) the polypeptide is synthesized but is proteolytically labile, or (4) the mutant protein is structurally intact but catalytically impaired. To differentiate among these possibilities, the HgCl,inducible translation products of both plasmids were examined via [3'S]cysteine/methionine pulse labeling. The resulting antoradiograph is shown in Figure 4. Upon induction with 0.5 pM HgCI,, both pMMA85 and pMMa558a559 clearly encode polypeptide bands of the correct molecular mass for nascent (9-kDa) and periplasmic (7.5-kDa) MerP, indicating that mer operon induction has occurred. [Jackson and Summen (1982) have shown that MerT is present at significantly lower levels than MerA and MerP and thus cannot be seen in this total cell protein gel.] However, while pMMa558a559 also encodes a distinct protein band of identical molecular mass to wild-type mercuric reductase (59 kDa), no such band can be seen, either visually or by scanning laser densitometry, from pMMA85

corresponding to the N-terminaldeleted or 'clipped" form of the enzyme (50 kDa). Clipped mercuric reductase retains 14 of the 18 cysteines and methionines found per subunit of the Alass8AlasS9protein, putting the lower detection limit for densitometry detection of clipped protein at ca. 6% of the level of the Ala,s8Alas59enzyme (data not shown). Lastly, over the time course of this experiment, there is no significant decrease in the intensity of the Ala,,,Ala,,, 59-kDa protein band, indicating that this mutant protein is not particularly sensitive to proteolysis. In summary, the above results show that the AlaSssAlaSS9 mutation must result in either an incorrectly folded protein or a catalytically incapacitated enzyme, while deletion of the codons for residues 2-86 of merA results in a very significant decrease in protein expression. Ala,,&aSs9 Mercuric Reductase Ouerproduction and Purijicotion. To facilitate purification, the Alas,,Alass9 mutant merA gene was cloned behind the lac promotor of plasmid pSE181 as shown in Figure 2. This construction, pMMOa558a559, differs from the wild-type and mutant mercuric reductase overproducing plasmids described by

N- and C-Terminal Cys Pairs of Mercuric Reductase Schultz et al. (1985) in that only the 1861 bp Aha11 fragment of pMMa558a559 (containing the Ala558Ala559 mutant merA gene and the beginning of merD) was inserted into the pSE181 vector. In contrast, pPSO1, the wild-type enzyme overproducing plasmid, was constructed by using a much larger, 3551 bp NarIIPuuII pJOE114 fragment as the insert into pSE181. pMMOa558a559 directs the overproduction of Ala558Ala559 mercuric reductase to >6.5% total soluble protein as determined by transhydrogenase activity in crude cell extracts (data not shown). For comparison, pPSOl directs the overproduction of wild-type mercuric reductase to 5% total soluble protein (Schultz et al., 1985). Ala558Ala559 mercuric reductase was isolated in > 100-mg quantities via a one-step purification using Orange A affinity chromatography. This purification scheme has been used previously for the wild-type protein as well as all of the redox-active disulfide mutants (Fox & Walsh, 1982; Schultz et al., 1985; Distefano et al., 1989). Since binding of the protein to Orange A and its elution with NADPH presumably depend on an intact nicotinamide binding site, the success of this affinity chromatography step implied that the Ala558Ala559 mutant protein is properly folded and is functional for NADP H binding. Thiol Titrations. The thiol content of KBr-dialyzed Ala558Ala559protein in the absence or presence of excess NADPH was determined with DTNB as described by Riddles et al. (1983). Upon denaturation with 6 M guanidine hydrochloride, 2.1 and 3.9 thiols were titrated, respectively. This increase of two thiols upon NADPH addition confirms the presence of an intact redox-active disulfide in the Ala558Ala559 enzyme. Titrations performed in 4-5 M guanidine hydrochloride, as described by Schultz et al. (1985) for the active site mutants, consistently resulted in a difference of ca. 1.3 thiols between oxidized and reduced Ala558Ala559 enzyme, possibly due to incomplete protein denaturation. When we denatured the wild-type enzyme in 6 M guanidine hydrochloride, 2.1 and 4.8 thiols were titrated in its oxidized and NADPH-reduced forms. This is consistent with both complete reduction of the redox-active disulfide and partial reduction of a second disulfide, as discussed in the following paper (Miller et al., 1989). Physical Properties. Ala558Ala559 mercuric reductase exhibits many of the same physical properties as the wild-type enzyme. Both Ala558Ala559 and wild-type proteins when freshly isolated migrate with an apparent subunit molecular mass of 59 kDa on 9% SDS-polyacrylamide gels. However, over time (1-4 weeks) both proteins become increasingly clipped, perhaps via a minimal protease contaminant, and migrate with an apparent molecular mass of 50 kDa. Fox and Walsh (1983) showed that this clipping results in the removal of an 85 amino acid N-terminal domain, but that clipped wild-type enzyme does not differ catalytically from unclipped enzyme. The native molecular masses of both the mutant and wildtype enzymes are somewhat ambiguous when determined by gel chromatography; they elute as single peaks between the elution volumes expected for dimeric and trimeric holoenzymes (Le., between 120 and 180 kDa). Evidence for both dimeric and trimeric mercuric reductases has been previously reported (Fox & Walsh, 1982; Rinderle et al., 1983; Schottel, 1978), although given its homology with the other disulfide oxidoreductases, it has been assumed that the biologically relevant structure is a dimer. In the present study, we observe that older enzyme samples that have become clipped behave more like dimeric species by gel chromatography, while unclipped samples behave more like trimers (data not shown). These results

Biochemistry, Vol. 28, No. 3, 1989

1189

suggest that mercuric reductase is indeed a dimer with the N-terminal domain making the protein less globular (S. Silver, personal communication), i.e., of a seemingly higher molecular mass, and are consistent with results of Rinderle et al. (1983), who observed that clipped samples of RlOO mercuric reductase also exhibit dimeric native molecular mass. The spectral ratio A272/A456 is 6.3 for both the wild-type and the Ala558Ala559 mutant, corresponding to a maximal flavin loading of 1 FAD per protein monomer (Fox & Walsh, 1983). In contrast to some of the redox-active disulfide mutants (Distefano et al., 1989), no significant flavin loss was observed in this mutant even after prolonged dialysis, indicating that there is little or no disruption of the FAD binding domain caused by the Ala558Ala559 mutation. Spectroscopic Properties and Electron Capacity. Like other oxidoreductases of this class, in the absence of nicotinamide adenine dinucleotides, wild-type mercuric reductase can exist in three spectrally distinct redox states (Fox & Walsh, 1982): (1) fully oxidized (E,, = oxidized flavin and redox-active disulfide), characterized by an absorbance maximum at 456 nm ( E = 11.3 mM-' cm-I), (2) two electron reduced (EH2 = oxidized flavin and redox-active dithiol), possessing a longwavelength charge-transfer absorbance from interaction of the Cys140thiolate with oxidized flavin, and (3) four electron or fully reduced (EH4 = reduced flavin and redox-active dithiol) that exhibits a bleached spectrum due to reduction of the FAD prosthetic group. Figure 5A shows a redox titration of the Ala55sAla559 mutant protein. From this figure it can be seen that, unlike the active site mutants described by Distefano et al. (1989) and Schultz et al. (1985), the visible spectra of the Ala558Ala559 enzyme are remarkably similar to those of the wild-type enzyme. [Wild-type spectra are not shown in Figure 5 for the sake of simplicity, but may be found in the following paper (Miller et al., 1989)l. In fact, the E,, spectra of the two are superimposable with the same absorbance maximum of 456 nm and extinction coefficient of 11.3 mM-' cm-'. Similarly, as the Ala558Ala559 enzyme is titrated with dithionite, it progresses through the same set of spectral changes as does the wild type, with their respective EH2 and EH4 spectra again superimposable. The EH2 flavin absorbance maximum for both is 438 nm (e = 8.6 mM-' cm-'), while the charge-transfer absorbance maximum for both is 540 nm (e = 2.6 mM-' cm-I). By plotting A,,, vs electron equivalents added (Figure 5A, inset), we see that full charge-transfer absorbance (EH,) develops after addition of 1.6electron equiv and that the enzyme is fully reduced to EH4 upon the input of another 2.3 electron equiv. We also note that the 2 electron equiv lag observed upon titration of the wild type before its oxidized spectrum begins to change (Fox & Walsh, 1983; Miller et al., 1989) is totally absent from the Ala558Ala559 titration. Thus, Ala558Ala559 mercuric reductase has a total redox capacity of 4,electrons. A titration of Ala558Ala559 mercuric reductase with NADPH, the enzyme's physiological reductant, is shown in Figure 5B. Much like its behavior upon titration with dithionite, we again see this mutant proceed through the same set of equilibrium charge-transfer species as the wild-type enzyme. The first of these is an EH2-NADP+ complex, which exhibits a blue-shifted flavin absorbance maximum at 448 nm ( E = 8.0 mM-' cm-') and a charge-transfer absorbance band centered around 560 nm ( E = 3.0 mM-' cm-I). Further addition of NADPH leads to the formation of an EH2-NADPH complex whose flavin and charge-transfer absorbances are centered at 444 nm (e = 8.6 mM-' cm-') and 530 nm ( E = 4.9 mM-' cm-I), respectively. As can be seen from the following paper

1190 Biochemistry. Vol. 28, No. 3, 1989

'22.W A V E L E N G M (nml

Dithionite and NADPH titrations of Ala55EAla559 mercuric reductase. KBr-dialyzed samples were made anaerobic and titrated with dithionite or NADPH as described under Methods. Spectra shown in (A) are of enzyme samples in the presence of (1) 0.0 (Eox), (2) 1.60 (EHz),and (3) 3.30 (EH,) equiv of dithionite. (Spectrum 3 contains some absorbance from reduced methylviologen, which was added to the titration as a mediator.) Part A insert follows charge-transfer absorption, as a function of equivalents of dithionite/enzyme monomer. Part B shows the spectra of enzyme in the presence of (1) 0.0 (Eox),(2) 1.3 equiv (EH2-NADP+),and (3) 5.0 equiv (EH2-NADPH) of NADPH, while the inset follows ASM (isosbestic point for the E,, to EHz-NADP+ transition) and A650 (indicativeof changes from Eo, to EH2-NADP+ to EH2-NADPH) as a function of equivalents of NADPH/enzyme monomer. FIGURE 5 :

(Miller et al., 1989), the spectra observed in this titration closely match those obtained with the wild-type enzyme. However, the wild-type and Ala55sAla559enzymes do differ significantly in the number of NADPH equivalents each requires for formation of the two complexes. Miller et al. show that, at 25 OC, equilibrium EH2-NADP+ formation by the wild-type enzyme requires 2 equiv of NADPH, while EHzNADPH is fully formed only when NADPH is present in greater than a 10 molar excess. In contrast, as shown in Figure 5B, inset, the Ala55sAla559 enzyme takes up only 1.1 equiv of NADPH to form a stable EH2-NADP+ complex, and EH2-NADPH formation is complete with only 3.3 total equiv of NADPH. The fact that the Ala55EAla559 mutation produces no significant perturbations in the visible absorbance spectra of the various oxidized and reduced mercuric reductase species indicates that, compared to C Y S ,and ~ ~CYS,,~, the C-terminal cysteines are not in intimate contact with the flavin. Yet, given the effect of this mutation on the enzyme's Hg(SR), reductase activity (see below), it seemed likely that CyssSsand/or Cys559 is located within the active site and may thus have an effect on the FAD fluorescence spectrum. However, the fluorescence properties of the E,, forms of both enzymes are quite similar, both having excitation and emission maxima at 4 O C of 455

Moore and Walsh and 520 nm, respectively (data not shown). In fact, the only significant differences in the fluorescence spectra of the two enzymes are their respective intensities when compared to that of free FAD in solution. These data are presented and discussed in detail in the following paper (Miller et al., 1989). Redox Potentials. We have previously reported E,,/EH2 and EH2/EH4 midpoint potentials for wild-type mercuric reductase to be -269 and -335 mV, respectively (Fox & Walsh, 1982). The E,,/EHz value was determined via dithionite titration in the presence of the redox dyes l-deazariboflavin and phenosafrarin, while the EH2/EH4couple was obtained by measuring the amount of disproportionation of EH2 to yield E,, and EH4 at various pH values. In the current work we have exploited a continuously reducing xanthine/xanthine oxidase system developed by Dr. Vincent Massey for determining midpoint potentials [see footnote 2, Distefano et al. (1989)]. From separate titrations using either 1-deazariboflavin (E" = -280 mV) or benzylviologen (EO' = -358 mV) as indicator dye, we determined the E,,/EH2 potential of the Ala55sAla55senzyme to be -310 or -292 mV, respectively (average = -300 mV). Plots of EOM vs log (E,,/EH,) for each case gave slopes of 29 and 3 1 mV, respectively, confirming the two-electron nature of this reduction. We also used benzylviologen as the indicator for the EH2/EH4 potential determination. Reduction from EHz to EH4, however, is clearly a complicated process as evidenced by the distinct biphasicity of the EoM vs log (EHz/EH4) plot (data not shown). The first half of the reduction [Le., log (EHz/EH4) > 01 yields a slope of 68 mV and midpoint potential of -348 mV, while the second half [log (EH2/EH4) < 01 has a slope of 32 mV and the midpoint potential of -35 1 mV. Such evidence for two categories of reduction sites was first observed by V. Massey with the C y ~ , ~ ~ Sand er,~~ Ser135Cys140 mutant enzymes and will be fully addressed in a later paper (V. Massey et al., unpublished results). Enzymatic Activities. We have assayed three aerobic and two anaerobic catalytic activities for the Ala55sAla559 and wild-type mercuric reductases. The aerobic activities are Hg(SR)2-independent NADPH oxidation (0, reduction), NADPH/thio-NADP' transhydrogenation, and Hg(SR),dependent NADPH oxidation. The anaerobic activities are DTNB reduction and Hg(SR)2 reduction via 203Hg(0)volatilization. The results of these assays are summarized in Tables I1 and 111. Except where noted, wild-type catalytic parameters were determined in parallel with those of the mutant both to serve as a control and to confirm previous results. All rates are expressed as the rate per enzyme monomer. 0, Reduction. In all activities tested not involving Hg(II), the Ala55sAlaj59enzyme is again remarkably similar to wildtype mercuric reductase, For example, under aerobic conditions both enzymes have a Hg(SR),-independent NADPH oxidase activity: with 200 yM NADPH and air-saturated buffer (37 "C), they both catalyze the production of H 2 0 zat a rate of 2 m i d . This rate was determined both by NADPH consumption (decrease of A340) and by 0, consumption (oxygen electrode). Schultz et al. (1985) have shown previously, via catalase addition, that the product of this activity is indeed H202. Transhydrogenation. Both wild-type and mutant enzymes also catalyze the transhydrogenation of thio-NADP' and NADPH. This activity allows the determination of a K , for NADPH (in 100 pM thio-NADP+), which is 0.7 1 pM for the Ala55sAlas59 enzyme and 0.8 WMfor the wild type (Schultz et al., 1985). The k,, for transhydrogenation (25 "C) is 116 min-l for the AlajjSAlaSS9mutant, representing a 2-fold rate

Biochemistry, Vol. 28, No. 3, 1989 1191

N- and C-Terminal Cys Pairs of Mercuric Reductase Table 11: Catalvtic Activities of Wild-Tvpe and Ala,saAla,,c Mercuric Reductases activity transhydrogenation

conditions and parameter monitored aerobic, 25 "C NADPH consumption

A1a558A1a559 K , = 0.71 pM NADPH k,, = 116 min-' NADPH oxidation/02 reduction aerobic, 37 OC O2 or NADPH consumption T N = 2.0 min-' DTNB reduction K , = 1.2 mM DTNB anaerobic, 25 O C TNB- production k,, = 132 min-' Ksi = 5.9 mM DTNB Hg(SR)&mulated NADPH oxidation aerobic, 37 OC NADPH consumption K, = 45 pM HgC12 k,, = 6.6 m i d Hg(SR)2-stimulated NADPH oxidation aerobic, 37 OC O2 consumption K , = 88 pM HgC12 k,, = 8.2 m i d "Fox & Walsh, 1982. bRates are expressed as turnover numbers (TN) if the rate was only examined under one set of conditions (Le., neither substrate concentration was varied). k,, indicates an apparent kat; Le., the concentration of only one substrate was varied, while the other was kept constant.

enhancement over that of wild-type enzyme. In this aspect the mutant resembles the Cy~,~~-alkylated wild-type enzyme, which has a 2.5-6-fold higher transhydrogenase k,,, than its nonalkylated form (Fox & Walsh, 1983). DTNB Reduction. Schultz et al. (1985) previously reported that the wild-type and Cys135Ser140mutant enzymes can catalytically reduce the aryl disulfide 5,5'-dithiobis(2-nitrobenzoate) (DTNB) in an activity analogous to the thiol/disulfide interchange reactions catalyzed by the other members of this enzyme family. They reported K , and k,,, values for the wild-type enzyme of 0.4 mM DTNB and 220 min-l, respectively. However, Au (1986) subsequently reported that the 220-min-' value for k,,, was not reproducible. Thus we decided to reexamine the reduction of DTNB by wild-type mercuric reductase, as well as determine kinetic parameters for the Ala5&laS5, mutant. All DTNB reduction assays were conducted in anaerobic solutions with 200 pM NADPH as described under Methods. Under these conditions, we have observed two previously unreported phenomena: (1) an NADPH- and DTNB-dependent enzyme activation and (2) a second-site DTNB-dependent inhibition. The activation results in biphasic kinetics with an initial slow TNB production phase being replaced by a faster linear rate over a period of 2-8 min. Both the amount of activation and the time required to reach the linear rate were dependent on [DTNB] with 5 mM and above giving a 4.5-fold activation after 6-8 min. The activation was not affected by prior incubation of the enzyme with either NADPH or DTNB alone and thus requires the presence of both substrates. The second phenomenon, that of a second inhibitory binding site for DTNB, is evidenced by a nonlinear [DTNB] vs [DTNB]/u plot above 5 mM DTNB. Using the faster, linear rates of TNB production described above, we have determined K,, k,,, and Ksi values (Cornish-Bowden, 1981) for the wild type to be 1.4 mM DTNB, 103 m i d , and 7.0 mM DTNB, respectively. The Ala558Ala559 enzyme is quite similar, with a K , of 1.2 mM DTNB, k,, of 132 min-', and Ksi of 5.9 mM DTNB. Aerobic Hg(SR)2-DependentNADPH Oxidation. While the Ala558Ala559 mutant appears to be very like the wild type in its catalysis of the above Hg(I1)-independent reactions, examination of its behavior upon addition of Hg(SR)2 reveals that it is quite different in its ability to handle and reduce Hg(I1). Under aerobic conditions, Fox and Walsh (1982) showed that the wild-type enzyme exhibits biphasic consumption of NADPH upon Hg(SR), addition, with an initial fast phase giving a k,, of 8 10 min-' and a slower, nearly linear phase giving a k,,, of 340 min-'. Using the faster phase, we found that for wild-type enzyme, the K , for Hg(1I) in the presence of 1 mM P-mercaptoethanol is 5.0 pM,consistent with the value of 12 pM reported by Fox and Walsh (1982). In contrast, under the same conditions the Ala558Ala55g mutant

wild type K , = 0.8 pM NADPH' k,, = 46 m i d TN = 2.0 min-l K , = 1.4 mM DTNB k,, = 102 min-' Ksi = 7.0 mM DTNB K , = 5.0 pM HgCl2 k,, = 8 10 min-' ND

Table 111: 2osHg Volatilization by Wild-Type and Ala558Aki559 Mercuric Reductases and Yeast Glutathione Reductase enzyme turnover turnovers (uM) no.' ( m i d ) re1 rate monitored wild type 0.002 318 19875 42900 AlaSS8AlaSS9 1 .O 0.28 18 16 0.5 0.30 19 134 yeast GR 4.0 0.016 1 9 0.035 1 .o 2 25 4Linear least-squares fit through all points above 60 p M [203Hg].

exhibits only a linear rate of NADPH consumption with a k,, of 6.6 min-' and K , of 45 pM Hg(SR)? Interestingly, Schultz and co-workers (1985) observed almost identical kinetic parameters (kat = 6.5 min-', K, = 48 pM) for the Ser13&ysl, mutant with Hg(CN)2. Recently, Distefano et al. (1989) have shown this NADPH consumption to result from a saturable increase in the NADPH-dependent 02-reduction rate rather than from Hg(I1) turnover. In light of this, we also examined the effects of Hg(SR)2 on the O2 consumption rate of the AlaSS8AlaSS9 mutant and likewise observed saturable kinetics with K,,, = 88 pM HgClz (in the presence of 1 mM p-mercaptoethanol) and kat = 8.2 mi&. Thus, within experimental error, essentially all of the Hg(SR),-dependent NADPH consumption observed for the Ala558Ala559 mutant under aerobic conditions can be accounted for by an Hg(I1)-effected increase in the oxidase activity. Anaerobic 203Hg(0)Volatilization. The above evidence indicates that although the Ala558Ala55g mutation drastically alters the enzyme's ability to reduce Hg(II), it has little effect on its other activities. In fact, one could argue that the Ala558Ala559 mutant, by virtue of its aryl disulfide (DTNB) reduction activity coupled with its inability to handle Hg(II), is actually a truer member of the flavin disulfide oxidoreductase family than is wild-type mercuric reductase. If this is indeed the case, then one might expect that if the Ala558Ala55g enzyme retains any residual Hg(SR), reducing ability, other members of this family might also have similar, adventitious levels of mercuric ion reductase activity. We have investigated the above possibility using a sensitive, anaerobic 203Hg(0)volatilization assay (Schottel, 1978; Distefano et al., 1989). With this assay, we tested the Hg(SR), reducing ability of the wild-type and AlaS58Ala559 mercuric reductases and yeast glutathione reductase. Typical volatilization curves for all three enzymes are shown in Figure 6, and the kinetic data are summarized in Table 111. We have found that both the Ala558Ala559 mutant mercuric reductase and yeast glutathione reductase can reduce Hg(II), albeit at very low rates. The former catalyzes Hg(0) evolution 1000fold slower than wild-type mercuric reductase while the latter is 10-fold slower still. That these slow volatilizations were

1192 Biochemistry, Vol, 28, No. 3, 1989

Moore and Walsh

100

80 m

._

c 2

60

0

100

200

300

400

500

600

time (min)

FIGURE6: M3Hg(0)volatilization curves for wild-type and Ala5s8Ala559 mercuric reductases and yeast glutathione reductase. Evolution of 203Hg(0)from anaerobic assay mixtures containing 400 pM NADPH, 1 mM cysteine, and 100 pM Zo3HgCl,was followed as described under Methods. The volatilization curves shown are those for assay mixtures containing no enzyme (0), 4.0 pM yeast glutathione reductase (+), 0.5 pM AlaS58AlaS59 mercuric reductase (A),and 2 nM wild-type mercuric reductase (m).

enzyme catalyzed was confirmed by showing that the rates changed with enzyme concentration (Table 111) and by following 203Hg(0)evolution from a control assay mixture lacking enzyme (Figure 6).

DISCUSSION The main goal of this work has been to identify unique structural features of mercuric ion reductase that enable it to carry out the last step in the bacterial mercury detoxification pathway, Le., reduction of Hg(I1) to Hg(0). We have approached this problem by using site-directed mutagenesis experiments specifically designed to address two standing hypotheses: (1) that the C-terminal cysteine pair, cys5&ys55,, thiols are involved in correct Hg(I1) positioning in the enzyme's active site and (2) that the N-terminal cysteine pair, C y ~ , ~ C y sthiols , ~ , participate in intracellular Hg(I1) transport (Brown et al., 1983). The first hypothesis was based on both molecular modeling studies and previous evidence that the C-terminal cysteine pair, cys55&s559 (Tn50J mercuric reductase has 56 1 residues total), has access to the enzyme's active site. Chemical modification experiments by Fox and Walsh (1 983) showed that if [ 14C]iodoacetamide,a sulfhydryl-specific labeling reagent, is incubated with oxidized mercuric reductase, no label is incorporated into the protein. However, if mercuric reductase is previously reduced with dithionite to the EH2 state, iodoacetamide is incorporated with a stoichiometry of 1 iodoacetamide/monomer, and 20% of this label is found in the C-terminal peptide. By modeling the amino acid sequence of mercuric reductase onto the 2.0-A crystal structure of glutathione reductase (no such structure currently exists for mercuric reductase), Brown et al. (1983) suggested that this cysteine pair may be involved in Hg(I1) binding at the active site of the opposite subunit. In the current study, we have constructed an Ala558Ala5S9 mutant mercuric reductase and have shown that it encodes a supersensitive phenotype in vivo and that its catalytic competence for Hg(I1) reduction in vitro has been reduced by ca. 1000-fold (Table 111). The fact that this mutant is so similar physically and spectrally to the wild-type enzyme and that it catalyzes NADPH oxidation, transhydrogenation, and DTNB

reduction with kinetic parameters so similar to those of wild type suggests that much of the active site residue alignment must be unchanged. The key observation that only Hg(I1) reduction is differentially affected singles out Cys558or Cys,,, or both as crucial components for either Hg(I1) coordination or electron transfer during catalysis. The necessity for the C-terminal cysteines is also brought out by the fact that the FAD/FADH2 redox potential for the Ala558AlaS59 mutant (see below) is actually more favorable for Hg(I1) reduction than is that of the wild type. We have also shown that another member of the flavin disulfide oxidoreductase family, yeast glutathione reductase, possesses a very low, but detectable, level of mercuric ion reductase activity. [This finding is consistent with results from the preceding paper (Distefano et al., 1989) showing that even FADH2 free in solution is capable of slowly reducing Hg(II).] It has long been proposed that mercuric reductase and glutathione reductase evolved from a common ancestral oxidoreductase [see Summers (1986)]. However, the key question of how substrate exclusivity developed between these two enzymes had not been previously addressed. The observation that removal of cys558 and C Y S ,from ~ ~ mercuric reductase results in a decrease in its Hg(I1) reduction rate by 3 orders of magnitude, while the rest of this enzyme molecule contributes only a IO-fold increase in the same catalytic rate over glutathione reductase, indicates that the C-terminal cysteines are a major part of the answer. In line with this, the following paper (Miller et al., 1989) presents data showing that the redox state of the C-terminal cysteines has a dramatic effect on the Hg(I1) reductase activity of the wild-type enzyme. Therein, hypotheses are presented for exactly how cys558 and Cys559 might participate in catalysis. Several lines of evidence have suggested that the N-terminal cysteine pair might be involved in a cytoplasmic Hg(I1)transport or -binding function: (1) the first 85 amino acids of mercuric reductase can be removed in vitro with no effect on catalytic activity (Fox & Walsh, 1983), yet this domain is conserved in three known mercuric reductases (Brown et al., 1983; Misra et al., 1985; Laddaga et al., 1987); (2) at the protein level, the N-terminal domain shows >30% sequence homology to MerP, the periplasmic Hg(I1)-binding protein; and (3) MerP contains only two cysteines, cys33and cys36, that are found in a region having considerable homology to the sequences surrounding the C y ~ , ~ C ymercuric s,~ reductase pair (Summers, 1986). The similarities between MerP and the N-terminal domain of mercuric reductase led Brown and Goddette (1984) to propose that merA may have arisen as the result of a gene duplication event that combined a primordial disulfide oxidoreductase gene with an ancestral merP. Brown (1985) has further hypothesized that mercuric reductase may transiently associate with the merT gene product at the cell membrane, with incoming Hg(I1) being transferred from MerT to the Cysl0Cysl3thiol pair. Under this model the Hg(I1) would then be transferred to the Cys558Cyss59pair where it would be correctly positioned, with the pair, for reduction. Thus the N-terminal domain of mercuric reductase would play a parallel role in the cytoplasm to that played by MerP in the periplasm. Recent site-directed mutagenesis studies of the MerP cysJ$ys36 pair have shown that removal of either residue does indeed disrupt the normal functioning of the mercury detoxification pathway (Nancy Hamlett, personal communication). Cys to Gly mutants exhibit a Hg(I1) resistance phenotype intermediate between wild-type resistance and background

N- and C-Terminal Cys Pairs of Mercuric Reductase sensitivity (Le., the resistance phenotype of cells containing no mer genes). These results are in agreement with those expected for a mer operon encoding a working reductase but a nonfunctional transport system. Concordantly, Lund and Brown (1987) have recently shown that deletion of the transport genes, leaving merA intact, results in a similar phenotype. From these observations, one might expect that if the CyslOCysl3pair does indeed perform a Hg(I1)-binding function as part of the transport system, Cys to Ala mutants would be less resistant to Hg(I1). However, under the conditions employed here, we could see no difference in the conferred mercury resistance of the Cysl&la13 and AlaloAla13mutants when compared to that of wild-type mer operon. Using a minimum inhibitory concentration assay, Brown and coworkers (personal communication) have been able to observe a small phenotypic difference between the C y ~ , ~ A mutant la~~ and the wild type: If the cells are preinduced with low concentrations of HgClz for 1-2 h and then tested for HgC12 resistance on agar plates, the CysIoAlal3cells are slightly less resistant than the wild-type cells; without the preinduction no difference was observed. Currently, we do not know whether this small difference is due to a transport aberration or to a decrease in the overall number of cellular thiols to which ligation of Hg(I1) is not deleterious. One possible cause for the lack of observable phenotypic change is the fact that all of our in vivo assays have been performed using multiple-copy plasmids. Under these circumstances, induction of the mer operon produces much more protein than is necessary for resistance (Nakahara et al., 1979a) and may thus swamp out any small Hg(I1)-transport perturbations. Ni’Bhriain et al. (1983) have previously noted a similar but opposite effect with Tn5 insertion mutations in the merD gene; Le., such insertions confer Hg(I1) sensitivity only when present on multiple-copy plasmids. Currently, Brown and wworkers are working to clone the Cysl&la13and AlaloAla13mutant merA genes onto single-copy plasmids to assess their in vivo effects under nonamplified conditions. Given that mutation of the N-terminal cysteine pair has little or no effect on mer operon function, we decided to address more stringently the role of the entire N-terminal domain by examining the consequences of its complete removal in vivo. This was accomplished by simply looping out the DNA coding for residues 2-86, making Ala8, the second amino acid in the coding region. This gene, merAA85, retains all of the upstream sequences of wild-type merA, including its SD sequence and spacing. However, as shown under Results, this “clipped” gene is not efficiently translated and thus results in a Hg(I1)-supersensitive phenotype. While we do not presently understand the reason for this phenomenon, a clue may be found in the sequence now immediately downstream from the ATG start codon. This sequence, ATGGCCGCCGCCGCC ..., with its high G C content, may effectively inhibit translation initiation [for review see Stormo (1986)l. Summary. We have constructed four mutations in mercuric ion reductase to probe the function of its N- and C-terminal cysteine pairs in Hg(I1) detoxification. In vivo resistance assays indicate that the N-terminal cysteines are not essential for HgC12 resistance while in vitro characterization of the Ala558Ala559 mutant enzyme has revealed a function for the C-terminal cysteines in catalysis. Additional studies with single Cys to Ala mutants of the C-terminal pair and/or hybrid mutants, e.g., a mixture of Ala135Ala140and Ala558Ala559 subunits, may help to address the questions of placement of the four thiol side chains during Hg(I1) reduction.

Biochemistry, Vol. 28, No. 3, 1989

1193

ACKNOWLEDGMENTS We thank A. Summers, S. Miller, V. Massey, C. Williams, and D. Ballou for providing helpful insights and for reading preliminary copies of the manuscript. We thank V. Massey for determining fluorescence spectra and for providing details of his yet unpublished xanthinelxanthine oxidase system for determination of redox potentials. We also gratefully acknowledge M. Distefano, K. Au, P. Schultz, and other members of the Walsh group for their contributions. Registry No. Cys, 52-90-4; Ala, 56-41-7; NADPH, 53-57-6; DTNB, 69-78-3; HgCl,, 7487-94-7; mercuric ion reductase, 6788093-7; mercury, 7439-97-6.

REFERENCES Amersham (1983) M13 Cloning and Sequencing Handbook, Amersham Corp., Arlington Heights, IL. Au, K. G. (1986) Mechanistic Studies of Active Site Mutants of Mercuric Reductase; Stereochemical Studies of a Fluoroacetate Halidohydrolase (Ph.D. thesis) MIT, Cambridge, MA. Biosearch (1 984) Saml Operator’s Manual, Biosearch, San Rafael, CA. Brown, N. L. (1985) Trends Biochem. Sci. 10, 400-403. Brown, N. L., & Goddette, D. (1984) in Flavins and Flavoproteins (Bray, R. C., Engel, P. C., & Mayhew, S. G., Eds.) pp 165-168, de Gruyter, Berlin. Brown, N. L., Ford, S. J., Pridmore, R. D., & Fritzinger, D. C. (1983) Biochemistry 22, 4089-4095. Casola, L., & Massey, V. (1966) J . Biol. Chem. 241, 498 5-4993. Cornish-Bowden, A. (198 1) Fundamentals of Enzyme Kinetics, pp 93-95, Butterworths, Boston, MA. Distefano, M. D., Au, K. G., & Walsh, C. T. (1989) Biochemistry (preceding paper in this issue). Foster, T. J. (1983) Microbiol. Rev. 47, 361-409. Fox, B., & Walsh, C. T. (1982) J . Biol. Chem. 257, 2498-2503. Fox, B. S . , & Walsh, C. T. (1983) Biochemistry 22, 4082-4088. Hames, B. D. (1 98 1) in Gel Electrophoresis of Proteins: A Practical Approach (Hames, B. D., & Rickwood, D., Eds.) IRL Press, Washington, DC. Hanahan, D. (1985) in DNA Cloning: A Practical Approach (Clover, D. M., Ed.) Vol. I, IRL Press Ltd., Oxford, England. Jackson, W. J., & Summers, A. 0. (1982) J . Bacteriol. 151, 962-910. Kunkel, T. A. (1985) Proc. Natl. Acad. Sci. U.S.A. 82, 488-492. Kunkel, T. A. (1988) Methods Enzymol. (in press). Laddaga, R. A., Chu, L., Misra, K. T., & Silver, S. (1987) Proc. Natl. Acad. Sci. U.S.A. 8 4 , 5106-5110. Lund, P. A., & Brown, N . L. (1987) Gene 52, 207-214. Maniatis, T., Fritsch, E. F., & Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Massey, V., & Williams, C. H., Jr. (1965) J . Biol. Chem. 240, 4470-4480. Messing, J. (1983) Methods Enzymol. 101, 20-79. Miller, S. M., Ballou, D. P., Massey, V., Williams, C. H., Jr., & Walsh, C. T. (1986) J . Biol. Chem. 261, 8081-8084. Miller, S . M., Moore, M. J., Massey, V., Williams, C. H., Jr., Distefano, M. D., Ballou, D. P., & Walsh, C. T. (1989) Biochemistry (following paper in this issue). Misra, T. K., Brown, N. L., Fritzinger, D. C., Pridmore, R. D., Barnes, W. M., Haberstroh, L., & Silver, S. (1984)

Biochemistry 1989, 28, 1 194-1 205

1194

Proc. Natl. Acad. Sci. U.S.A.81, 5975-5979. Misra, T. K., Brown, N. L., Haberstroh, L., Schmidt, A., Goddette, D., & Silver, S. (1985) Gene 34, 253-262. Nakahara, H., Kinscherf, T. G., Silver, S., Miki, T., Easton, A. M., & Rownd, R. H. (1979a) J.Bacteriol. 138, 284-287. Nakahara, H., Silver, S., Miki, T., & Rownd, R. H. (1979b) J. Bacteriol. 140, 161-166. Ni'Bhriain, N. N., Silver, S., & Foster, T. L. (1983) J. Bacteriol. 155, 690-703. O'Halloran, T. V., & Walsh, C. T. (1987) Science 235, 21 1-214. Riddles, P. W., Blakeley, R. L., & Zerner, B. (1983) Methods Enzymol. 91, 49-60. Rinderle, S. J., Booth, J. E., & Williams, J. W. (1983) Biochemistry 22, 869-876. Schottel, J. L. (1978) J. Biol. Chem. 253, 4341-4349. Schultz, P. G., Au, K. G., & Walsh, C. T. (1985) Biochemistry 24, 6840-6848.

Seidman, M. (1985) BMBiochemica 2 ( 5 ) , 10. Shames, S . L., Fairlamb, A. H., Cerami, A., & Walsh, C. T. (1986) Biochemistry 25, 3519-3526. Stanisich, V. A., Bennett, P. M., & Richmond, M. H. (1977) J. Bacteriol. 129, 1227-1233. Stormo, G. D. (1986) in Maximizing Gene Expression (Reznikoff, W., & Gold, L., Eds.) pp 195-224, Butterworths, Stoneham, MA. Summers, A. 0. (1986) Annu. Rev. Microbiol. 40, 607-634. Walsh, C., Fisher, J., Spencer, R., Graham, D. W., Ashton, W. T., Brown, J. E., Brown, R. D., & Rogers, E. F. (1978) Biochemistry 17, 1942-195 1. Williams, C. H., Jr., Arscott, D. L., Matthews, R. G., Thorpe, C., & Wilkinson, K. D. (1979) Methods Enzymol. 6 2 0 , 185-198. Wilson, G. S . (1978) Methods Enzymol. 54, 396-410. Zoller, M. J., & Smith, M. (1983) Methods Enzymol. 100, 468-500.

Evidence for the Participation of Cys558and Cys559at the Active Site of Mercuric Reductase? Susan M . Miller,*,* Melissa J. Moore,# Vincent Massey,* Charles H. Williams, Jr.,*," Mark D. Distefano,l David P. Ballou,* and Christopher T. Walshs Department of Biological Chemistry, University of Michigan, Ann Arbor, Michigan 481 09, Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, Massachusetts 021 15, and Veterans Administration Medical Center, Ann Arbor, Michigan 48105 Received June 1 , 1988: Revised Manuscript Received September 20, 1988 ABSTRACT: Mercuric reductase, with FAD and a reducible disulfide a t the active site, catalyzes the two-

electron reduction of Hg(I1) by NADPH. Addition of reducing equivalents rapidly produces a spectrally distinct E H 2 form of the enzyme containing oxidized FAD and reduced active site thiols. Formation of E H 2 has previously been reported to require only 2 electrons for reduction of the active site disulfide. W e present results of anaerobic titrations of mercuric reductase with N A D P H and dithionite showing that the equilibrium conversion of oxidized enzyme to EH2actually requires 2 equiv of reducing agent or 4 electrons. Kinetic studies conducted both at 4 O C and at 25 "C indicate that reduction of the active site occurs rapidly, as previously reported [Sahlman, L., & Lindskog, S. (1983) Biochem. Biophys. Res. Commun. 117,231-2371; this is followed by a slower reduction of another redox group via reaction with the active site. Thiol titrations of denatured E,, and EH2 enzyme forms show that an additional disulfide is the group in communication with the active site. [ 14C]Iodoacetamide labeling experiments demonstrate that the C-terminal residues, cysss8 and Cysss9, are involved in this disulfide. The fluorescence, but not the absorbance, of the enzyme-bound FAD was found to be highly dependent on the redox state of the C-terminal thiols. Thus, E,, with cysss8 and Cysssg as thiols exhibits less than 50% of the fluorescence of E,, where these residues are present as a disulfide, indicating that the thiols remain intimately associated with the active site. Initial velocity measurements show that the auxiliary disulfide must be reduced before catalytic Hg(I1) reduction can occur, consistent with the report of a preactivation phenomenon with N A D P H or cysteine [Sandstrom, A,, & Lindskog, S . (1987) Eur. J. Biochem. 164, 243-2491. A modified minimal catalytic mechanism is proposed as well as several chemical mechanisms for the Hg(I1) reduction step. M e r c u r i c reductase ( M R ) ~plays a crucial role in bacterial detoxification of mercurials as it catalyzes the 2e- reduction of Hg(I1) shown in eq 1. The enzyme exhibits extensive Hg(SR)2 + NADPH + H + Hgo + NADP' 2RSH (1)

-

+

'This work was supported by National Institutes of Health Grants G M 11 106 (V.M.), G M 21444 (C.H.W.), G M 20877 (D.P.B.), and G M 21643 (C.T.W.). M.J.M. is an N S F predoctoral fellow. * To whom correspondence should be addressed. !University of Michigan. Harvard Medical School. '1 Veterans Administration Medical Center.

0006-2960/89/0428- 1 194$01.50/0

similarities to the pyridine nucleotide disulfide oxidoreductases, lipoamide dehydrogenase and glutathione reductase, both in primary sequence and in spectral properties. Thus, they all

'

Abbreviations: DTNB, 5,5'-dithiobis(2-nitrobenzoic acid); DTT, dithiothreitol; E,,, mercuric reductase containing oxidized FAD and oxidized active site thiols, but reduced C-terminal thiols; EHz, mercuric reductase containing oxidized FAD and the active site thiols reduced; EH4, mercuric reductase containing FADHz and reduced active site thiols; E,,, mercuric reductase containing oxidized FAD and active site disulfide; HPLC, high-performance liquid chromatography; MR, mercuric reductase; 4-PDS, 4,4'-dithiodipyridine; TPCK, N-tosyl-~-phenylalanine chloromethyl ketone.

0 1989 American Chemical Society